Advances in pharmaceutical development, especially in the field of therapeutic antibodies, are rapidly enabling and/or improving the treatment of many diseases. These advances, by reaching novel target spaces and providing novel mechanisms of action are increasingly improving the quality of lives of patients even with the most severe and challenging diseases. One challenge for the health care system in general and patients in particular is that the costs of new drugs, enabled by of these pharmaceutical advances, are also rapidly increasing. The high costs are a result of the investments required for the development of pharmaceuticals, especially of antibodies, which currently exceed one billion dollars per marketed product. The high risk of failure in development and very long developmental timelines make these investments inevitable. It may take over fifteen years from the time of identification of a potential therapeutic antibody until it reaches the market and can benefit patients. Each stage of development, from identification, pre-clinical, clinical to market entry is riddled with challenges and risks. Pharmaceutical companies are constantly working to reduce developmental costs by reducing timelines and risks of failure in order to get the most effective medicines into the hands of patients quickly.
The following disclosure provides a valuable advance which allows for faster identification of the optimal therapeutic antibodies for the treatment of any disease. Therapeutic antibody candidates must fulfill a number of development criteria in order to make it to the market, such as, long term stability, low aggregation propensity and high expression yields. The disclosed advance increases the probability and speed of identifying an antibody that can fulfill all of the rigorous development criteria right from the start. The resultant antibody will be less expensive to produce and will be effective and safe in the treatment of numerous diseases.
A well known method of identifying therapeutic antibodies is through the use of phage display technology. Phage display utilizes virus-like particles that are grown in bacteria to display antibodies. One benefit of this technology is that the libraries used are massive, with up to 1×1011  antibodies, which can quickly be tested for binding to any target relevant for any disease. See, for example, Knappik et al., (2000), “Fully synthetic human combinatorial antibody libraries (HuCAL) based on modular consensus frameworks and CDRs randomized with trinucleotides,” J. Mol. Biol. 11; 296(1):57-86, and U.S. Pat. No. 6,300,064, both of which are incorporated by reference in their entireties. The benefit of working with such large numbers is that the output of a screening against a target may result in hundreds of antibodies that bind to the therapeutic target, all of which could be therapeutically relevant. A problem, though, is that often only a few of these antibodies are developable, meaning that they can meet all of the rigorous criteria required in order to make it to the market.
In order for a new phage display collection to shorten the identification timelines and reduce the inherent risks, the collection should comprise antibodies having the properties which are necessary for selection and clinical development and which will result in safe and effective treatment in patients. Such properties include: 1) high phage display rates, so that each and every antibody of the collection can be tested against the target of interest; 2) high expression levels in both Fab and IgG1 formats, so that the antibody or fragment can be reproduced efficiently with the needed quantity; 3) high thermal stability in both Fab and IgG1 formats, to ensure structural and functional integrity of the molecules delivered to patients; 4) high stability in serum in both Fab and IgG1 formats, so that the antibody shows increased half-life and prolonged activity; 5) high monomeric content (% monomer) as determined by size exclusion chromatography (SEC) in both Fab and IgG1 formats as this signifies a low aggregation propensity; 6) high isoelectric point (pI) in IgG1 format; 7) high thermal stability in Fab and IgG1 formats before and after exposure to acid; 8) low turbidity in Fab or IgG1 formats before and after exposure to acid; 9) stable molecular radius and % polydispersity before and after exposure to acid; 10) low risk of immunogenicity, thereby increasing safety, and/or 11) high diversity, so that one collection can be used to identify many antibodies against any therapeutic target.
A collection, which in essential ways imitates the human immune system, should be highly valuable, or even the optimal solution. The human immune system is composed of antibodies encoded by germline genes. Antibodies, in part, comprise of a variable heavy chain and variable light chains. There are approximately 50 variable heavy chain germline genes and approximately 50 variable light chain germline genes, combined providing about 2,500 combinations of different variable heavy and light chain pairs. In humans, all 2500 of these combinations are believed to be produced. It has been found, though, that certain variable heavy chains, variable light chains and/or variable heavy and light chain combinations (pairs) are present at a higher level than others. It was hypothesized that there must be some reason that some are present more than others, and if so, that the highly present germline genes may have favorable functional properties. Therefore, one way of providing a collection of antibodies having favorable functional properties is to generate a collection comprising the abundant variable heavy chain, variable light chain, and/or variable heavy chain and variable light chain pairs present in the human immune repertoire.
In addition, the germline gene sequences present in humans are thought to have very low immunogenicity, for obvious reasons, therefore these sequences can be imitated in recombinant antibodies in order to lower the risk of immunogenicity.
Approaches to evaluate the variable heavy and light chain germline gene pairings prevalent in the human immune repertoire have been undertaken. See de Wildt et al., Analysis of heavy and light chain pairings indicates that receptor editing shapes the human antibody repertoire, J Mol Biol. 22; 285(3):895-901 (January 1999), which is incorporated by reference in its entirety. Wildt et al. took blood samples from human donors, sorted the IgG+ B cells, which had undergone somatic hypermutation, PCR amplified the cDNAs, sequenced each cDNA, and aligned each sequence to the known human variable domain germline genes. Wildt et al. observed that only a few germline genes dominated the immune repertoire and that the frequent heavy and light chain gene segments are often paired.
Attempts at maintaining the heavy and light chain variable domain pairings of individual B cells have also been undertaken. For example, libraries of variable domain “cognate pairs” have been disclosed. See Meijer et al., Isolation of human antibody repertoires with preservation of the natural heavy and light chain pairing, J Mol Biol., 358(3):764-72 (May 5, 2006); and WO2005042774, which are both incorporated by reference in their entirety. Libraries according to the techniques described in Meijer et al. have been generated from individual B cells from an immunized host. Generally, the B cells are sorted by FACS so that CD38HI B cells, which represent somatically hypermutated cells, are selected, their cDNAs are PCR amplified, and the antibody gene products are inserted into Fab vectors for selection. Such cognate pair libraries are not without their limitations. For example, the hosts providing the B cells typically are immunized; and the B cell populations sorted have been hypermutated, therefore, the resulting libraries are biased towards a particular immunogen.
Additionally, attempts at utilizing prominent variable heavy chain or variable light chains for collection generation have been undertaken. For example, in Shi et al., “De Novo Selection of High-Affinity Antibodies from Synthetic Fab Libraries Displayed on Phage as pIX Fusion Proteins; J Mol Biol., 397(2):385-96 (Mar. 26, 2010) and the respective patent application WO2009085462; and WO2006014498, which are incorporated by reference in their entireties. There, variable heavy chain or variable light chain germline protein sequences were incorporated into libraries based upon their frequency of use in the human immune repertoire.
Additional attempts have also been undertaken, which incorporate a specific germline pair into a collection. For example, WO1999020749, which is incorporated by reference in its entirety, describes a collection where its members comprise heavy chains having the canonical structure of a hypervariable loop encoded by the human germline heavy chain gene segment DP-47 (IGHV3-23) and/or framework regions encoded by the germline gene, and/or light chains having the canonical structure of a hypervariable loop encoded by the human germline light chain gene segment O2/O12 (IGKV1-39/1D-39) and/or framework regions encoded by the germline gene.
Additional approaches have generated libraries directly from or derived from B cells. For example, Glanville et al., Precise Determination of the Diversity of a Combinatorial Antibody Library Gives Insight into the Human Immunoglobulin Repertoire, Proc Natl Aced Sci 1; 106(48):20216-21 (December 2009), which is incorporated by reference in its entirety, which describes an antibody collection built from the diversity of 654 human donor Immunoglobulin M (IgM) repertoires. Specifically, the heavy and light chain V-gene cDNAs from 654 human donors were separately PCR amplified (separating the variable heavy and light chain pair) and the heavy and light chain domains were then randomly re-associated. WO2003052416, which is incorporated by reference in its entirety, also describes the isolation of B cells from a host exhibiting a pronounced response to a pathogen of interest, resulting from either an infection by a micro-organism or treatment with a vaccine. In WO2003052416, the cDNA encoding the CDR3 region of the variable regions was sequenced and antibody fragments comprising the dominant CDR3s were designed. WO2009100896, which is incorporated by reference in its entirety, describes the isolation of B cells from an immunized host, where the cDNAs encoding the variable heavy and light chain regions were sequenced and the abundance of the unparied variable heavy and variable light chain sequences was determined. In WO2009100896, libraries were synthesized comprising the randomly recombined variable heavy and variable light chains, wherein the antibodies were specific for one immunogen. A summary of these and additional approaches is found in Fuh et al., Synthetic antibodies as therapeutics, Expert Opin Biol Ther., 7(1):73-87 (January 2007), which is incorporated by reference in its entirety.
There is, therefore, a high need for a collection of antibodies or fragments thereof that incorporate the variable heavy and variable light chain gene pairs present in the human immune repertoire that have favorable biophysical properties relevant to development, while at the same time excluding the pairs that exist in nature, but do not have such biophysical properties. These and other needs are satisfied by the present invention.